UCL Genetics Institute


Seminar 26 November, Paul O'Reilly - Increasing statistical power in GWAS

8 November 2013

Seminar organised by the Bloomsbury Centre for Genetic Epidemiology and Statistics

Tuesday 26 November 2013, 4:30 - 5:30pm, Location: Taviton (16) 431, 14-16 Taviton Street (UCL)
Dr. Paul O'Reilly, Senior Lecturer, Social Genetic & Developmental Psychiatry, King's College London
Title: "Increasing statistical power in GWAS"

I will discuss three ways in which the statistical power to detect genetic variants in genome-wide association studies (GWAS) may be increased: (1) via consideration of the disease (inheritance) model, (2) through the choice of phenotype transformation, (3) by testing multiple related phenotypes simultaneously. We find that the usual approaches of employing an additive disease model, performing the log transformation, and testing phenotypes separately in GWAS often lead to substantially lower power to discover genetic variants affecting diseases and human traits than easily implemented alternatives.